Decoder with Nilay Patel

How Silicon Valley enshittified the internet

October 30, 2025

Key Takeaways Copied to clipboard!

  • Enshittification is a three-stage platform decay process where companies first attract users, then exploit business customers, and finally harvest all remaining surplus from both, leaving a 'pile of shit' platform. 
  • The acceleration of enshittification in tech is attributed to the removal of key disciplinary forces, specifically the collapse of interoperability due to restrictive IP laws (like DMCA 1201) and the erosion of worker power. 
  • The current Generative AI boom is economically unsustainable based on current revenue models, and for creative workers, the U.S. Copyright Office's stance that AI-generated works lack copyright offers a unique, boss-opposed opportunity via sectoral bargaining to increase creator compensation. 
  • Effective regulation of online platforms must focus on administrability, suggesting identity portability (like phone number portability) as a superior remedy to content moderation disputes for empowering users to leave toxic platforms. 
  • Labor organizing among tech workers is crucial for preventing the degradation of their working conditions to the level of warehouse or factory workers once they are no longer scarce. 
  • Despite conventional political science wisdom, global antitrust enforcement is currently gaining momentum across various jurisdictions (US, Canada, EU, UK, Asia), suggesting a repeal of the 'law of political gravity' that previously protected economic elites. 

Segments

Defining Enshittification Term
Copied to clipboard!
(00:04:11)
  • Key Takeaway: Cory Doctorow coined ’enshittification’ after a frustrating experience with TripAdvisor, using it as a framing device for platform decay.
  • Summary: The term ’enshittification’ was coined by Cory Doctorow to describe platform decay, stemming from a personal incident with TripAdvisor’s excessive trackers. It functions as a rallying cry for those analyzing why internet products are worsening. The core concept describes a causal narrative of platform decay.
Three Stages of Platform Decay
Copied to clipboard!
(00:05:59)
  • Key Takeaway: Platforms follow a pattern: first pleasing users to lock them in, then worsening service for users to benefit business customers, and finally worsening service for business customers to harvest all surplus for shareholders.
  • Summary: Stage one involves attracting and locking in end-users. Stage two occurs when platforms make things worse for users to benefit business customers who are also subsequently locked in. Stage three involves harvesting all remaining surplus from business customers, leaving only a minimal residue of value for users.
W3C and DRM Turning Point
Copied to clipboard!
(00:07:51)
  • Key Takeaway: The 2017 decision by the W3C to incorporate Digital Rights Management (DRM) into browsers, pressured by streaming companies, marked a critical turning point where external discipline on tech collapsed.
  • Summary: The W3C capitulated to movie studios and tech companies by adding DRM to browsers, threatening the web’s viability otherwise. This move made it illegal to modify browsers for lawful purposes like accessibility checks or security research. This event was highly demoralizing as a public interest institution was cornered into serving corporate interests.
Facebook as Iconic Enshittification Case
Copied to clipboard!
(00:10:32)
  • Key Takeaway: Facebook is the reference case for enshittification because its decay occurred under the same leadership (Mark Zuckerberg) since its inception, proving the process is structural, not dependent on founder succession.
  • Summary: Facebook initially lured users from MySpace by promising not to spy on them, locking users in via the collective action problem of social networks. Stage two involved abusing advertisers with surveillance data and publishers with traffic funnel dependency. The end stage is exemplified by the pivot to the metaverse, where all available value has been withdrawn.
Tech Discipline vs. Walmart
Copied to clipboard!
(00:17:01)
  • Key Takeaway: Tech enshittification is distinct from general capitalism because tech lacked the traditional market and regulatory discipline, further compounded by the loss of interoperability and worker leverage.
  • Summary: The growth of monopolies is directly linked to the drawdown of anti-monopoly law enforcement, evidenced by Zuckerberg’s written admission of buying Instagram to reduce competition. Tech had unique disciplinary checks—interoperability and scarce, productive workers—which were removed by IP laws and mass layoffs, respectively. The resulting monopsony power allows firms to dictate terms to suppliers and workers.
IP Law and Interoperability Loss
Copied to clipboard!
(00:19:55)
  • Key Takeaway: Anti-circumvention laws, particularly Section 1201 of the DMCA, eliminated interoperability, making it legally risky for users to install third-party software like ad blockers on apps, thus enabling greater abuse.
  • Summary: The universal nature of computers allows programmers to create software workarounds for platform decay, but IP laws have removed this check. Apps, unlike websites, are protected by IP that criminalizes installing ad blockers, allowing invasive practices like Chamberlain discontinuing HomeKit support to force users into an ad-filled proprietary app experience. This loss of interoperability is a key tech-specific mechanism for enshittification.
AI’s Role and Legal Frameworks
Copied to clipboard!
(00:30:03)
  • Key Takeaway: The legal foundations that enabled tech monopolies—limitations on intermediary liability (CDA 230) and solidified fair use (Google Books)—are the same foundations that currently permit large-scale AI training.
  • Summary: The excitement for disintermediation in the 90s was about avoiding gatekeepers, but the failure to police intermediary mergers created the current winner-take-all market. The lawfulness of scraping for indexing and counting copyrighted works for analysis is crucial for AI training, even if the resulting models are not human-authored. The WGA’s success in securing sectoral bargaining rights offers a better path for creators against AI than relying on expanded copyright law.
Economic Reality of Foundation Models
Copied to clipboard!
(00:51:36)
  • Key Takeaway: The current foundation model AI sector is economically unsound, characterized by massive capital expenditure ($700B+) far exceeding annual gross revenue ($45B), suggesting the bubble is likely to collapse.
  • Summary: The high compute demands of current models mean that intelligence is not scaling with word guessing; the models cannot maintain context for complex tasks like software engineering or radiology review. The projected revenue model relies on displacing high-wage jobs, but the total wage bill of affected sectors like illustration is negligible compared to training costs. The economic foundation is so weak that the number of proprietary foundation models could realistically drop to zero.
Sponsor Readout: Bedrock Robotics
Copied to clipboard!
(01:00:47)
  • Key Takeaway: Bedrock Robotics is using AWS Generative AI Accelerator support to bring advanced autonomy to heavy equipment to address the construction labor shortage.
  • Summary: Bedrock Robotics is developing AI for the built world to tackle America’s construction crisis, where half a million jobs remain unfilled. Participation in the 2024 AWS Gen AI Accelerator Program provided infrastructure support, leading to an $80 million funding raise a year later. The company focuses on physical AI, collaborating with partners like Amazon to positively affect the world.
Sponsor Readout: Carvana
Copied to clipboard!
(01:02:20)
  • Key Takeaway: Carvana offers a seven-day return policy for purchased vehicles, allowing customers to test the car after delivery or pickup.
  • Summary: The segment features a brief endorsement of Carvana’s process, emphasizing that customers receive exactly what they ordered. Customers can choose delivery or pickup and have one week to decide whether to keep the car or return it. Fees and exclusions may apply to the return policy.
Prescriptive Solutions to Enshittification
Copied to clipboard!
(01:02:53)
  • Key Takeaway: The core regulatory challenge in tech is administrability, making identity portability a more effective solution than defining and policing subjective harms like hate speech.
  • Summary: Regulating hate speech is difficult because it requires multi-year processes to define terms and adjudicate negligence, which is mismatched against harassment occurring thousands of times per second. A better approach is making it easier for users to leave platforms, similar to phone number portability. Modern networks like Mastodon support identity portability via small text files containing followers and blocks, which could be mandated for legacy silos like Twitter.
Labor Organizing and Worker Power
Copied to clipboard!
(01:06:46)
  • Key Takeaway: Tech workers missed the opportunity to organize when they were scarce, and without fear of organized labor, corporations will inevitably treat them like low-wage warehouse workers.
  • Summary: The historical precedent shows that capital sued for labor peace after violent conflicts, resulting in the NLRA as a compromise, but enforcement has favored limiting worker power. The current lack of a quorum at the National Labor Relations Board means unfair labor practices cannot be adjudicated, returning the environment to pre-legalization organizing conditions. Workers must now organize without state support, leveraging the historical praxis of unionization from a century ago.
Hope vs. Optimism in Activism
Copied to clipboard!
(01:09:27)
  • Key Takeaway: Hope, the belief that action can lead to improvement, is preferred over optimism, which suggests improvement will happen regardless of effort.
  • Summary: The speaker defines optimism as a form of fatalism, believing things will get better without action, whereas hope requires active participation. Activists can only see the next few steps toward a goal, but each successful step expands the visible terrain for future action. This moment is ripe for change due to widespread anger against oligarchy and unprecedented global antitrust actions against major tech firms.